You would expect 2 sequences of pseudo-randomly allocated numbers with different seeds to have a very low correlation coefficients, either side of zero. But not zero, because, well, randomness.

Here is the results from correlating 5 sequences of a million psuedo-random values each. Each sequence is generated with a different randomly chosen seed value:

All the coefficient values are small, so it looks like **Random** is doing a fair job. There are noticeably more negative than positive values. This could be an artefact of the pseudo-random generator or it could be randomness. You would need to do a lot more correlations to find out which. Either way, I think it is good enough for the purposes it is likely to be used for.

big-correl.transform (5.5 KB)

Note that we used the **Sequence** transform to create a millions rows from:

NB/ You will need to load the latest v1.45.1 snapshot (or later) to get the **Correlate** transform.